I was captain of the red team with Corey, Andrew, Alison, and Joel.
We had 4 days.
During the blueprint challenge and the beginning of our time in the shop, we thought it was a simple race to see who could go up and down the fastest. At this point, we were focusing on very simple designs that could move very quickly - mostly big buckets that caught all the water. Maybe it was when started asking Alison how many g's of acceleration she thought she could handle, but at some point production felt the need to inform us that it wasn't a race. They called me and Gui together for a quick meeting where they explained that they were looking for "elevator-like" speeds and that so long as it moved at a decent pace, we needn't go faster. We, of course, asked how they would determine the winner, and they told us it would be judged, making this the most subjective challenge so far.
For pretty much any design, it made sense to have a trough right at the top to capture all of the water and turn a "natural waterfall" into 28 feet of head. Sluice gates are a good choice for controlling flow because they can easily open and close under pressure and allow for big gates with lots of flow.
The main design I presented was a series of buckets on vertical loop. They would fill with water and the weight of the water would provide the motive force. I added a few details/refinements such as figuring out how big the buckets should be (1 large one massing a bit more than the elevator, and several small ones totaling a bit less) and exactly where valves needed to be and how to control them.
I also drew a secondary design much like what we actually built. Because of time and/or money constraints, the massive bucket loop structure might be too much to build. In this case, we could get most of the performance by putting a waterwheel near the bottom. This would allow gravity to convert the potential energy (head) into kinetic energy (speed), which could then all be used by a relatively compact waterwheel, which would be simpler to manufacture than the larger bucket system.
This was a 4-day build that started on Thanksgiving Day and ended on Sunday. We only worked a half day on Thursday, then worked 14 hour days (instead of the usual 12) the rest of the build to make up for it. Fortunately, we had no need for long team meetings back at the house after the extra-long shop days on this build. The holiday and weekend also made sourcing parts more difficult than usual. Thankfully, we mostly only needed lots of steel, some stuff from McMaster, and some stuff scavenged from previous builds, so I thought build 4 was actually harder in terms of sourcing parts. Production also helped us out a little bit by pre-buying a dozen or so different sizes of steel, which they gave us a list of. If we wanted any of those pieces, we could get them within an hour or two (at normal price).
Thanksgiving evening was a good time. First, and maybe most importantly, the cameras left. Then, they gave us a pretty good traditional Thanksgiving feast. Then we had a TV with 3 DVD's to watch (including The Avengers!). Also, we got kinda tipsy and had a pillow fight. It was good to just be friends for a few hours rather than competitors. Also, we had a little handheld video camera and we amused ourselves by filming the shenanigans as well as interviewing each other and parodying the style of questions we normally got asked as well as the "first-person" and "present tense" requirements.
The biggest cause of failed builds so far was running out of time. Additionally, since it was difficult to predict the performance (torque and RPM) of our wheel, we needed to be able to test it before we could pick the spool diameter (essentially a gear ratio). Therefore, I pushed my team to work quickly. I was constantly telling people to stop doing things the right way and just get it done. Everybody always wanted to seal everything perfectly and get rid of any leaks; I figured that so long as we were leaking less than 1 gallong/second, it really didn't matter given the size of our water source. Fortunately, I had a great team and everybody took the pushing good naturedly. I especially appreciate Alison and Joel's (who were already eliminated) willingness to work hard during the extra long 14 hour days in the shop. The nominal plan was to have the water wheel turning by the end of the 2nd day (without control and without connection to the elevator) so that we could measure its output and correctly size the spool. I never really thought we'd make this goal (we didn't), but I think it was useful to feel behind schedule to keep us moving quickly. As it was, we did get finished in time to test and improve things when the test didn't go well.
Once we learned it wasn't a race, I think we were all pretty quickly on board with the waterwheel plan. There was some effort to either find an off-the-shelf turbine or an off-the-shelf pump that we could run backwards as a turbine. Either would be more efficient (and require less work) than the waterwheel, but we couldn't find anything adequate in our price range that we could source in time.
As usual, we got 2 hours in the shop on the demonstration day (Wednesday). Almost the first thing we did, was order a barrel since it seemed like it would be handy no matter what design we used and we knew we couldn't get anything besides the steel production had available on Thursday. Somehow, we got the barrel Thursday morning; I think McMaster opened up for us in the morning. We cut the barrel in half, then attached the two halves end to end to build a trough to sit at the top and catch all of the water. At one end (the left), was a sluice gate that could be opened to allow water to flow into an attached box. Our (mine and Andrew's) calculations indicated that two big PVC pipes would give our wheel plenty of power to lift the elevator, but we made the box big enough for 4 pipes just in case. We attached 4 flanges, but left the caps on the two we weren't using. This bit of contingency planning, which I insisted on, ended up saving us later. Alison did most of the fabrication for this system with significant help from Joel on the reinforcing frame. I think Andrew did most of the triangular cantilevered stand that it sat on.
We went with a 4 foot diameter wheel because that was the biggest sheet metal we had easily available. Andrew and I (mostly Andrew) designed the shape of the vanes by modifying a drawing in some ancient paper we found. Then I got to bend them by hand (i.e. by hammer) because the shop had no sheet metal tools. We also had to roll the inner hoop by hand, foot, and hammer. Unfortunately, this left us with something less than round which had to be forced into place to seal against the inside of each vane. This required me getting inside the hoop and holding it in place with my foot (my arms were nowhere near strong enough) while Corey welded it (this is shown on the show).
Andrew and I designed the mount to have the pillow blocks holding the main axle suspended from the triangular cantilevered stand. This was slightly advantageous in terms of getting the wheel exactly where we wanted it. Unfortunately, we first mounted our triangular stand to the waterfall truss, then mount the wheel to the stand. With the underhung design the (quite heavy) wheel had to be held in place while it was bolted on. The funny thing is that while Andrew and I designed it this way, it was Corey, Joel, and Alison who had to deal with it on site. I jokingly refer to this as good leadership when I'm not apologizing for it.
We used disc brake from a motorcycle to slow and stop our wheel. This was one component that made me nervous since I've never worked with them before and we had a bad experience with home-made brakes on challenge 1. Joel has some experience with motorcycles, so he took the lead on this, sourcing it from a junkyard and adapting it onto our system.
The wheel, brake, and a spool are all on one large shaft. When the wheel turns, it winds a cable around our spool, which lifts the elevator. When there is no water on the wheel, the weight of the elevator causes it to unspool and turn backwards (if not braked).
The only controls were the sluice gate at the top and the brake at the bottom. Both could be quickly and easily converted to manual operation via rope (and maybe pulley) if the primary plan didn't pan out, but we preferred to use electronic control to make it more elevator-like. We used 3 linear actuators scavenged from previous challenges: 1 to push the master cylinder on the brake (via a lever), and 2 in series (for 12" throw) to open and close the sluice gate. We also have an optical encoder on the primary shaft to measure the motion of the elevator. I'm actually rather impressed that the encoder, protected by only a plastic bag, survived all of our tests in the middle of a waterfall. This ran into the electronics board from the previous challenge scavenged off the back of our robot. I simply removed the extra components and reused it (with some extra water-proofing). Alison made a nice control panel with three doorbell buttons for the 1st (bottom), 2nd (halfway up), and 3rd (top) floor.
The code was very simple - I think I wrote it in an hour at a folding table and chair on site while the rest of the team was rigging stuff. When an elevator button was pressed, if the elevator wasn't already there, it would go to the correct floor. To go up, it simply opened the sluice, then released the brake after a short pause to let the flow build up. To stop, just apply the brake. To go down, close the sluice, then release the brake after pausing to let the flow stop. The one cool feature was proportional speed control during the descent to produce a smooth, slow motion.
We were told we were allowed to use battery power as part of the control system so long as it wasn't providing motive force, but we preferred to be completely self-sufficient. To this end, we added a smaller (18") water wheel in the spilloff from the trough. We tried to use a car or motorcycle alternator to generate power off of it, but after two failed attempts (1 where we convinced ourselves the junkyard-sourced alternator was faulty, and 1 where we realized we simply couldn't get high enough RPM, even with gearing), we eventually used the BAM (a large DC motor) from the previous challenge as a generator. At the speed we turned it (probably about 100 RPM from the wheel through a 1:3 gearing up), it put out about 1.5 V, which was not enough to charge our 12V batteries. So we made a secondary supply out of 18 1.5V batteries (instead of the 2 12V's), which we could charge one at a time. We never had time to actually test running off of this supply, but I believe we could have (we had very low current requirements). We did have a pretty sweet charging sign that Alison made - it looked really sweet at night when we were testing it, though it wasn't too visible in direct sunlight.
We got the parts for this wheel waterjetted by WET, so it was one of the few pieces we actually made CAD for. Corey's CAD for the little wheel is actually used in the renderings on the show to represent the big wheel.
Rigging our device to the waterfall truss was a big part of the build and required working away (about short drive) from the shop. Knowing that rigging would take longer than expected, we started on that as soon as possible: on the 2nd day. We rigged our components as we finished them and work was prioritized to get some parts completely finished and ready for rigging rather than getting everything partially done. As team leader, I felt constrained to stay in the shop and oversee things there, so I put Corey in charge of rigging and had him run activites on site. I don't think I got out there at all until the final day.
On the last day, each team had a van dedicated to them to get to and from the test site, and for us, there was a lot of back and forth. The entertaining thing was that we all had microphones, so they knew when we would need the van. After a while, if the van was already at the test site and somebody had to get from the shop to the test site, they would just start talking to themselves about it a few minutes before they wanted to go. When they walked out the door, the van would be there.
When we first tested the full system, it didn't work, primarily due to an enormous amount of friction in the belay system. For safety, the elevators were belayed by a rope system not under our control. We knew about it, but expected it to have little effect until we tested and found out it had a ton of friction (maybe up to 100 lbs or so), possibly caused by crappy pulleys. I think on our first test, the elevator (with a sandbag Alison-mass-simulator) went up, but would not come back down because of the excessive friction. Then we added another sandbag, but it wouldn't go up between the extra friction and the less-than-expected wheel torque. Fortunately, we still had time to make changes:
By the end of this, we had it working reliably and had done full tests with Alison on board about 3 times the day before the actual competition. It was fortunate for us how far behind the other team was running. It meant that we had total freedom to turn the waterfall on and off as we wished, making it much easier to run our tests and make our changes since you can't really turn the waterfall on while anybody is trying to rig stuff.
Andrew and I did an overly simplified analysis through which we convinced ourselves that nozzles at the end of the PVC pipes wouldn't help. Essentially, we believed that gravity would turn all of the head (potential energy) into velocity for us without having to add nozzles. What we missed was that there were a ton of losses involved in the water flowing down the pipes, which slowed it down significantly. If we'd added nozzles, they would have restricted the flow, turning the head (potential energy) into pressure, which the nozzles would have converted into velocity with fewer losses than just falling down the pipes. If we'd added nozzles, we'd have gotten a lower flow rate, but it would have exited the pipes at a higher velocity, and we could have always compensated for the reduced flow rate by adding more pipes as necessary.
When the original deadline passed, our elevator was only sort of working. It was still very finicky and needed some more tweaking. However, the other team hadn't shown up yet, and we were told we had a few more hours to work. We used the time to tweak and test until we were confident, then left. We were later told we could work more on it in the morning, but decided there wasn't anything useful to do, so we slept in instead. When they finally brought is in somewhat before noon, the other team was still rigging. We sat around watching them, speculating about what wouldn't work, and generally being amused by their troubles. Eventually, we were sent over to the tent (shown during the Blue Team's post-test argument), where we couldn't watch. They finished rigging mid-afternoon, most of a day after the deadline.
While waiting for the other team to finish rigging, we all got pretty bored, including Kal Penn, and the judges. I think we hit Kal's trailer with a frisbee, after which he came out to see who was knocking, and we had to sheepishly explain what happened. They also all came over to hang out and chat for a bit. We met Kal's dog. I shared a peanut-butter-and-jelly sandwich with Kal. He showed us his trailer when somebody asked about it. The guest judge, Dezso Molnar also came by to chat for a while. This was probably the most time we had to hang out with the judges and Kal the entire time.
Alison gets all the credit for this. She asked me if we could get outfits. I told her we'd wait to the end to make sure it fit in our budget and that our stuff was going to work (I didn't want us in costume if it was going to fail). When we got near the end and things looked good, I gave her a budget and she made it happen.
Back (Eric's TBBT Home)